Single-cell gene expression analyses hold promise to characterize cellular heterogeneity, but current methods sacrifice either the coverage, sensitivity or throughput. Several methods exist for full-length cDNA construction from large amounts of RNA, including cap enrichment procedures (Maruyama, K. & Sugano, S., Gene 138, 171-174 (1994); Carninci, P. & Hayashizaki, Y., Meth. Enzymol. 303, 19-44 (1999); Das, M., et al., Physiol. Genomics 6, 57-80 (2001)), but it is still challenging to obtain full-length coverage from single-cell amounts of RNA. Existing methods use either 3′ end polyA-tailing of cDNA (e.g., Tang, F. et al., Nat. Methods 6, 377-382 (2009); Sasagawa, Y. et al., Genome Biol. 14, R31 (2013)) or template switching (Zhu, Y. Y., et al., BioTechniques 30, 892-897 (2001); Ramsköld, D. et al., Nat. Biotechnol. 30, 777-782 (2012)), whereas other methods sacrifice full-length coverage altogether for early multiplexing (Islam, S. et al., Genome Res. (2011). doi:10.1101/gr.110882.110; Hashimshony, T., et al., Cell Rep. 2, 666-673 (2012)). It has recently been shown that Smart-Seq, which relies on template switching, has more even read coverage across transcripts than polyA-tailing methods (Ramsköld, D. et al., Nat Biotechnol. 30, 777-782 (2012)), consistent with the common use of template switching in applications designed to directly capture RNA 5′ ends, including nanoCAGE (Plessy, C. et al., Nat. Methods 7, 528-534 (2010)) and STRT (Islam, S. et al., Genome Res. (2011). doi:10.1101/gr.110882.110). Single-cell applications utilizing template switching are dependent upon the efficiency of the reverse transcription, the template switching reaction, and a uniform polymerase chain reaction (PCR) preamplification to obtain representative cDNA in sufficient amounts for sequencing. Despite the widespread use of these reactions, no systematic efforts to improve cDNA library yield and average length from single-cell amounts have been reported.